Research map / Label-efficient and robust learning / Test-time adaptation
The Cone of Silence: Speech Separation by Localization
Teerapat Jenrungrot, Vivek Jayaram, Steve Seitz, Ira Kemelmacher-Shlizerman
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use 44.1 kHz instead of 16 kHz sampling rate (ablation) | Cone of Silence binary search · Synthetic separation | median SI-SDRi (dB): 14.132 -> 17.059 | Improved |
| use 44.1 kHz instead of 16 kHz sampling rate (ablation) | Cone of Silence · Synthetic localization, 2 voices + 1 BG | median angular error (deg): 3.98° -> 3.73° | Improved |
| give oracle source locations instead of binary search localization (ablation) | Cone of Silence · Synthetic separation | median SI-SDRi (dB): 17.059 -> 17.636 | Improved |
| pass window-size conditioning h to every encoder and decoder block rather than once (ablation) | Cone of Silence · Synthetic separation | separation quality | Improved |
| binary search instead of linear sweep | Cone of Silence · 2 voices + BG at 2 degree resolution | forward passes: 180 -> 32.64 | Improved |
| train with 50% real and 50% synthetic mixtures (ablation) | Cone of Silence · Real recorded data | separation performance | Improved |
| add more speakers than seen in training (2 to 8) | Cone of Silence · Synthetic | median SI-SDRi (dB): 13.9 -> 6.3 | Hurt |
Closest papers by meaning
- Latent Gaussian Activity Propagation: Using Smoothness and Structure to Separate and Localize Sounds in Large Noisy Environments (NeurIPS 2018 · similarity 0.87)
- Separate and Reconstruct: Asymmetric Encoder-Decoder for Speech Separation (NeurIPS 2024 · similarity 0.85)
- ArrayDPS: Unsupervised Blind Speech Separation with a Diffusion Prior (ICML 2025 · similarity 0.84)
- RealMAN: A Real-Recorded and Annotated Microphone Array Dataset for Dynamic Speech Enhancement and Localization (NeurIPS 2024 · similarity 0.84)
- ZeroSep: Separate Anything in Audio with Zero Training (NeurIPS 2025 · similarity 0.84)
- Object-aware Sound Source Localization via Audio-Visual Scene Understanding (CVPR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
